Cluster analysis is a means of grouping records based
upon attributes that make them similar. If plotted geometrically, the objects within the
clusters will be close together, while the distance between clusters will be further
apart. What makes this type of analysis different than regression analysis is the absence
of a "dependent variable". In a regression, a set of attributes are being mathematically
related to a variable we wish to predict, or make some inference upon. In cluster analysis,
there is no dependent variable. The attributes are related only to themselves and not to
any prediction variable.
In business, clustering is often used for marketing purposes. Clusters
are designed to group accounts according to similar characteristics so a proper marketing
campaign may be developed for each group. Cluster analysis is more of an art than a science
since decisions have to be made on the number of clusters and the interpretation which can
best describe the grouping. SAS and S-Plus software have a number of popular clustering
algorithms which can be used for marketing applications.